welcome back
Claude Code · Session 2

Skills, Agents
& Sub-agents.

Today you'll learn to hand your work off — so Claude picks it up itself.

Claude Code for entrepreneurswith Emma & Derek
Where we left off

Last time

In the Q&A, someone asked about building a whole group of agents — dynamic agents. So today we zoom in on one thing: what an agent really is.

2 / 30
First, the foundation

What is a large language model?

A program trained on an enormous amount of text. It learns patterns — then predicts what comes next, so well it feels like it understands you. Claude is an LLM.

“It doesn't look things up — it predicts what's most likely to come next.”

Predicting

Give it “the sky is …” and it fills in “blue”. Next word, next word — that's the whole trick.

Clustering

It groups things that mean the same — so “invoice”, “quote” and “receipt” sit close together.

3 / 30
You've probably done this already

You already talk to AI

Tools like ChatGPT and Gemini are LLMs too. You type a message — it types back.

That's the part everyone knows: a conversation. Today we go one step further.

✳ChatGPT ✦Gemini
What's a good name for my studio?
A few ideas: Atelier Nova, Studio Bright, Form & Field…
4 / 30
Same idea, different window

Chat vs. Claude Code

ChatGPT · Gemini — a chat box
Write a welcome text for my studio.
Sure! Here's a warm welcome text you can copy…

It tells you. You copy-paste.

Claude Code — your terminal
claude
› Add a welcome text to my site
· reading your brand files…
· editing index.html…
✓ done — opening preview

It does it — right in your files.

Same conversation — but now it happens in your terminal. That's the switch we're making today.

5 / 30
The whole point

An LLM can chat.
An agent can BUILD.

That's the leap: from words on a screen to work that's actually done — in your files.

6 / 30
Under the hood

The agentic loop

The moment you ask, the same loop runs — watch it live on the right. — click through it →

Your prompt“Build my landing page”
↓
agentic loop
Gather context
↓
Take action
↓
Verify results
repeats
↓
Done
claude — live
›
●
●
●
↻
✓
›
⎋

While it works, you can keep adding context — just type. Want it to stop? Press Esc.

Source: how-claude-code-works7 / 30
Zoom out · how it fits

What powers that loop?

Claude Code · the harness
Models
reasoning · the brain
Tools
acting · the hands
→
The agentic loop
look → act → check ↻

Models reason, tools act. The harness binds them together and runs the loop. That's how it fits.

8 / 30
The tools themselves

Claude's tools

Files · read, write, create, organize
e.g. update your quote in the Word file
Search · find things in your files
e.g. find where your prices are
Run · run commands, test, deploy
e.g. put your website online
Web · look things up, docs, error messages
e.g. look up a supplier's opening hours
Code insight · spotting errors in code
advanced — we'll get to this later

Each tool step feeds the next — the agentic loop in action. The details come over the next weeks, module by module.

Source: how-claude-code-works9 / 30
The term: agentic harness

Claude Code = the harness

A frame around the model that gives it tools, memory and a workspace. That's what turns a language model into a capable agent.

Claude Code · the harness tools memory workspace
Claudethe model · LLM
Source: how-claude-code-works10 / 30
Behind the scenes

Look under the hood

AI doesn't come with a manual — and it's an ongoing product, still growing every week.

To really grasp what it can do, it helps to look under the hood. The engineers behind Claude Code work in one place: the .claude directory.

You don't have to know it all. We'll touch the parts that matter: tools · hooks · skills · agents · subagents · workflows.

~/.claude/
├─ CLAUDE.md
├─ settings.json · tools & permissions
├─ skills/
├─ agents/ · subagents
├─ hooks/
├─ commands/
└─ workflows/
11 / 30
Behind the scenes

What happens in your .claude folder

You ask: “Build a landing page.” Here's what happens under the hood:

01
Reads your rules · CLAUDE.md
02
Grabs the right skill · skills/landing-page
03
Uses tools and writes the files · files + run
04
Checks & opens — done · verify → done

That folder looks overwhelming. But most of it is engine room you skip past — and this is exactly what we open up over the coming weeks.

12 / 30
Back to CLAUDE.md

CLAUDE.md = the memory

Like setting an alarm or picking a wallpaper on your phone: this is where you save your rules and preferences. Claude reads it every time you start a session.

So it knows straight away how to work with you.

Write it once. Every conversation, it remembers.

13 / 30
There's no right way

“There is no one correct way to use Claude Code.”

Boris Cherny — co-creator of Claude Code at Anthropic — says his own CLAUDE.md is “surprisingly vanilla”. It works fine out of the box. You can tweak and hack it however you like. But you don't have to.

So: no stress about doing it “wrong”. There is no wrong.

Boris Cherny · @bcherny14 / 30
Remember what an LLM is

What a language model is great at

Write a welcome text for my studio.
“Welcome to Studio Bright — where bold ideas take shape.”
Summarize this long email.
In short: they'd like a quote by Friday.
Group these 20 ideas into themes.
Three themes: brand · product · launch.
✓ Writing text — posts, pages, replies.
✓ Summaries — long → short.
✓ Clustering ideas — grouping what belongs together.

Predict the next word · cluster meaning. But there's one thing it's bad at →

15 / 30
The one weak spot

Procedural knowledge

Procedural knowledge is knowing the steps to get something done — the recipe, in the right order.

1
2
3
4
5
?
— loses the thread

An LLM starts a long list fine — then drifts. It predicts the next word; it was never built to hold a whole procedure in its head.

Great at language, weak at following a long list of steps. That's exactly the gap a skill fills.

16 / 30
So how do we solve this?

This is where a skill comes in

Without a skill you spell out every step, every time. A skill lets you write those steps once — then just call it.

Without a skill — you write it all, every time
› Build a landing page for my studio.
  Colours: ink + electric blue only.
  Font Helvetica, no spaced-caps.
  Tone: warm, plain, never salesy.
  Sections: hero, offer, proof, contact.
  Logo top-left, links top-right.
  Mobile-friendly. Show a preview.
With a skill — you just call it
› /landing-page
✓ same result — every step, on brand.

A skill is just a recipe — your steps, saved once — so a short call gives the full result.

17 / 30
What a skill actually does

Same prompt. Two very different results.

Without the skill
claude
›
your-site.com

Welcome

We help you grow your business.

Get started
With the /smartgirls skill
claudesmartgirls skill
›
your-site.com
welcome to

SmartGirls

Warm, editorial, unmistakably yours — cream & bordeaux, elegant serif, every single time.

Join the club

Same request — but the skill already knows your style, so the right one comes out on-brand.

18 / 30
The heart of today · you do it yourself

Skills — your own recipe

Pick one recurring task from your business. You ask once — Claude writes the skill file you'll reuse forever.

1 · You ask
“Make a skill called brand-guide that captures my brand style: colours [x], font [y], tone [z]. Write down the rules.”
2 · Claude writes the skill
brand-guide.md
---
name: brand-guide
description: My brand style — colours, font, tone
---
1. Colours: ink + electric blue only
2. Font: Helvetica — no spaced-caps
3. Tone: warm, plain, no jargon
4. Never: clutter, stock photos, grey-on-grey
5. Check every page & deck against this

Goal: you go home today with one Skill you'll actually use this week.

19 / 30
From skills to agents

How do skills relate to agents?

An agent is a helper with a role that decides for itself which steps to take. A skill is a recipe it can reach for — so the agent doesn't guess the procedure, it picks the right skill and runs it.

Skill

Your recipe — the steps and preferences, written down once.

Agent

A helper that decides on its own when to reach for that recipe.

Skills are the building blocks. Agents are what pick them up and run — which is exactly why the skill has to be good first.

20 / 30
Why we start here

Skills first — on purpose

An agent makes decisions for itself. That's powerful — but first you want to be able to trust it.

1 · You decide

With a skill, you choose when to run it.

2 · You iterate

You use it, tweak it, until the output is reliable.

3 · Then you trust it

So when the agent later picks it on its own, you already know the result is right.

Build the recipe first. Hand it to the agent second.

21 / 30
Today

Three pieces of gear

Each one gives you time back. No tech — just handing work off.

1 · you, now
Skills
Capture a recurring task once. Then call it with a single sentence.
← you build this one
2 · demo
Agents
A fixed helper with a role: your copywriter, your brand-style checker.
3 · demo
Sub-agents
Several helpers at once, each on its own piece of the work.
22 / 30
From one worker to a team

One agent, or an agent with staff?

Same worker either way — it runs the same loop of observe → think → act. What changes is the role. Think of a kitchen.

The food truck · one agent

One person does every step themselves — one pair of hands, tasks wait in line.

The professional kitchen · parallel
chef = lead agent

A lead agent splits the work across subagents who cook in parallel.

Same worker — different role. Alone at the top → the agent; hired by another → a subagent.

23 / 30
How they relate

Skill · Agent · Subagent

Agent
A helper with a role. Decides when to use a skill.
your marketing agent
delegates a sub-task
returns a summary
· its own room ·
Subagent
The same — an agent, but hired by the agent. Own memory.
code reviewer
both reach for it
Skill
A recipe — the steps, written once.
/brand-guide  /invoice

The agent hands a sub-task to a subagent — and both reach for the same skills below. One recipe, shared by whoever needs it.

24 / 30
Going one level deeper

Building your own agent

The essentials in one line: one clear role, a precise trigger, the right tools — and you iterate until you trust it.

Today — the idea

What an agent is, how it wakes up (you · a schedule · an event), and why skills come first.

In the module — the how

Step by step: write the role, set the trigger, pick the tools, and ship a working agent of your own.

We keep it simple here. The full build-along lives in the module on the website.

25 / 30
Today · the agent

Agents — how to use them

You make an agent once: you give it a role, a task and your rules. After that you call it and it works on its own — you don't have to spell out every step.

Brand-style checker

Reviews every page against your colours, font and tone — and says what's off.

Steady copywriter

Writes in your voice: product copy, newsletters, posts.

Quote maker

Grabs your details, fills in the quote and checks the total.

Important: an agent remembers its role and sticks to your rules. You can have several — each with its own job.

26 / 30
How it comes together

An agent with skills

Your skills are the recipes. Your agent is the cook with a fixed role. Give the agent access to your skills, and it reaches for the right recipe itself.

You trigger it and give a goal →
Marketing agent
Decides which skill — and when.
picks the
right one
landing-page
newsletter
brand-check

You give the agent a goal — it decides which skill to use, and when. That's why the output holds up, even when you're not watching.

27 / 30
Next week

Dynamic agents & sub-agents

Today you made one helper. Next week we go further: agents that decide for themselves which steps are needed, and sub-agents — several helpers at once, each on its own piece of the work.

One for the copy, one for the design, one for the structure — in parallel.

Then we'll also explain how sub-agents work. For now: good to know it exists.

28 / 30
Take home · this week

Build a skill that works — then hand it to an agent.

Two steps: make one Skill you trust, then give it to an agent and tell it when to reach for that skill on its own.

Post your result on the forum.29 / 30
How we do this together

Questions along the way?

See you Friday

Claude Code for entrepreneurs30 / 30
← → navigate · F fullscreen
EN NL